The artificial intelligence trade is no longer being supported only by expectations about what the technology might eventually produce. It is increasingly showing up in corporate earnings.

But the gap between AI-linked companies and the rest of the market may be approaching a more difficult test.

With most S&P 500 companies having reported second-quarter results, Bank of America analysts found median earnings-per-share growth of 28% among AI-linked companies, compared with 12% for companies outside that group, according to a Reuters market briefing Monday. Consensus estimates cited by the bank expect AI-linked earnings growth to slow to 16% next quarter.

That still represents substantial growth. The question is whether the market is beginning to move from rewarding AI exposure broadly to distinguishing between companies that can turn spending into durable revenue and those still relying on the promise of future demand.

AI has become an earnings category

The scale of the divergence is notable because it changes the nature of the market argument around AI.

For much of the current cycle, investors were asked to price enormous capital expenditures against forecasts of future productivity, cloud demand and software adoption. The central question was whether spending on chips, data centers and AI infrastructure would eventually generate enough revenue to justify the investment.

Now there is more evidence that parts of the ecosystem are doing exactly that.

Memory suppliers, networking companies, data-center equipment providers, semiconductor manufacturers and cloud infrastructure businesses have all benefited from the physical build-out required to train and run increasingly large AI systems. The earnings effect has also extended beyond companies that are typically labeled as pure technology plays.

That is helping support broader market expectations. MarketWatch reported Monday that JPMorgan strategists raised their S&P 500 earnings expectations after a strong reporting season, pointing in part to improving evidence that AI monetization may begin catching up with spending.

That is a meaningful shift from the question that dominated 2024 and 2025: When will the return arrive?

The next quarter may be less forgiving

The market is now confronting a different problem. High growth creates a difficult comparison.

If AI-linked earnings slow from a median 28% to something closer to the 16% consensus expectation cited by Bank of America, the group can remain fundamentally strong while still disappointing investors accustomed to accelerating results.

The distinction matters because valuations have already absorbed a great deal of optimism.

Strong earnings can support high prices. They do not eliminate the risk that investors have priced in even stronger earnings.

That is why the next wave of reports matters. Applied Materials, Cisco and CoreWeave are among the AI-exposed companies scheduled to report this week, giving investors another look at demand for semiconductors, networking capacity and cloud infrastructure.

The results will help answer whether the current earnings advantage is broadening, holding steady or beginning to normalize.

The market is starting to separate AI users from AI beneficiaries

Nearly every large company now has an AI narrative. Far fewer have an AI earnings story.

That distinction is likely to become more important as adoption spreads.

A retailer using generative AI to improve customer service is not economically equivalent to a company selling the networking equipment required to build an AI cluster. A software company adding an assistant is not necessarily capturing the same economics as a cloud provider selling the compute underneath it.

Even within the infrastructure layer, the economics vary considerably. Some companies benefit from immediate capacity constraints and contracted demand. Others face enormous capital requirements and a longer path to positive free cash flow.

The next phase of the AI market may therefore be less about whether a company is exposed to AI and more about where it sits in the value chain.

From narrative to operating performance

There is a larger corporate lesson in the earnings divide.

AI has moved far enough into the economy that executives can increasingly be judged on operating outcomes rather than adoption announcements. Investors can ask whether AI is expanding margins, creating new revenue, lowering costs, improving productivity or producing demand for an existing product.

That is healthier than treating every reference to AI as equivalent.

The technology can still transform industries without every AI-related stock growing at the same rate indefinitely.

In fact, the narrowing of the earnings gap may eventually be evidence of something more mature: AI becoming embedded throughout the economy rather than concentrated in a small set of obvious winners.

For now, however, the numbers still show a market divided between companies capturing unusually strong AI-linked growth and everyone else.

The next question is how long that premium lasts.